Two Contributions of Constraint Programming to Machine Learning

Abstract

A constraint is a relation with an active behavior. For a given relation, we propose to learn a representation adapted to this active behavior. It yields two contributions. The first is a generic meta-technique for classifier improvement showing performances comparable to boosting. The second lies in the ability of using the learned concept in constraint-based decision or optimization problems. It opens a new way of integrating Machine Learning in Decision Support Systems.

Cite

Text

Lallouet and Legtchenko. "Two Contributions of Constraint Programming to Machine Learning." European Conference on Machine Learning, 2005. doi:10.1007/11564096_61

Markdown

[Lallouet and Legtchenko. "Two Contributions of Constraint Programming to Machine Learning." European Conference on Machine Learning, 2005.](https://mlanthology.org/ecmlpkdd/2005/lallouet2005ecml-two/) doi:10.1007/11564096_61

BibTeX

@inproceedings{lallouet2005ecml-two,
  title     = {{Two Contributions of Constraint Programming to Machine Learning}},
  author    = {Lallouet, Arnaud and Legtchenko, Andrei},
  booktitle = {European Conference on Machine Learning},
  year      = {2005},
  pages     = {617-624},
  doi       = {10.1007/11564096_61},
  url       = {https://mlanthology.org/ecmlpkdd/2005/lallouet2005ecml-two/}
}